Details
-
Bug
-
Resolution: Cannot Reproduce
-
Major
-
7.1.0
-
7.1.0-1975
-
Untriaged
-
Centos 64-bit
-
-
1
-
No
-
CX Sprint 277, CX Sprint 278
Description
Testcase name -
test_cbas_node_crash_while_swap_rebalancing_cbas_nodes_does_not_change_actual_number_of_replicas
Steps to reproduce -
- Have a 3 node cluster with following services -
- Node 1 - KV, N1ql, index
- Node 2 - CBAS
- Node 3 - CBAS
- Create datasets, dataverses, indexes and synonyms.
- set cbas replica to 2.
- Now swap rebalance cbas nodes.
- Stop couchbase-server on incoming node, such that rebalance started in step 3 fails.
- restart the couchbase-server on incoming node.
- rebalance again and wait for rebalance to finish (now we have 3 cbas node, so 2 replicas are now supported)
- disconnect all the local links
- mark one of the CBAS node as failed over, such that replica becomes active.
- Now verify the doc count in the datasets created in step 2.
Few Docs in one of the datasets is missing.
This issue is intermittent.